Calculating Dot Gain....Murray Davies Equation?

[mk162]

does anybody have one in an excel spreadsheet or a working link online? It's a confusing equation and figuring it by hand is taking too long.

If you couldn't tell we are really trying to nail down our dot gain curves...so far I am pretty pleased, just looking to get it closer.

[Printficient]

I think Joe Clarke has one.  I'll ask him on Monday.

[Rockers]

does anybody have one in an excel spreadsheet or a working link online? It's a confusing equation and figuring it by hand is taking too long.

If you couldn't tell we are really trying to nail down our dot gain curves...so far I am pretty pleased, just looking to get it closer.

The dot gain adjustments, where do I take care of them? In Photoshop or in my Wasatch Softrip? This is still new terrain for us. So if someone could elaborate a little bit more on this I would greatly appreciate it.

[Screen Dan]

The dot gain adjustments, where do I take care of them? In Photoshop or in my Wasatch Softrip? This is still new terrain for us. So if someone could elaborate a little bit more on this I would greatly appreciate it.

I'm not positive but I'm pretty sure that if your rip pays attention to color profiles you can take care of it in Photoshop.

I take care of it in our rip.  We use FusionRIP, but I would be shocked if every modern RIP package didn't have a way to take care of this.

[Get Shirts]

Accurip will also handle this for you.

[Dottonedan]

I should answer your question first as best I can.  I can't tell you how to do it "by hand".  I don't understand that part.
Are you meaning "manually" ?  If so, you still can't "accurately" since you have no way of determining if a 30% is actually a 30%. There is no math without having your actual printed readings. You know where you are starting, but you must know where you end, to see how many steps you must walk back.


You can only figure out the compensation (once you have your results). From there, it's a very simple process.

PHOTOSHOP CURVES:
Photoshop does handle dot gain (in curves) that is used to adjust individual channels per individual file. An artist may manually adjust for dot gain in many areas before output.  These are arbitrary adjustments often based on experience and the knowledge of ink types and substrate such as, is it a sweat shirt or a white tee?.

PHOTOSHOP COLOR SETTINGS:
Photoshop has dot gain adjustments (for conversion) from RGB to CMYK and as it pertains to grayscale in "color Setup" using your Color settings and imputing your ink type, common gain adjustment/compensation etc.

Both of the above are still affected by or compounded by any (output to film) compensation via any RIP. I stress that, because of the fact that any compensation done before the RIP will be adding to, or interfering with any predetermined or arbitrary settings. The RIP will adjust "outside of" or in addition to the Photoshop settings. 

It's also important to mention that most compensation (should be) compensated for in your RIP if it's a good detailed RIP. The reason it's more important to apply compensation here, is that "compensation" should be based off of compensation/control "results" from ...

A, your imaging device being used to print the image to the screen (or film).

B, your actual press capabilities, settings and habits.

It should not typically be based off of what program you print it from.

Using compensation via a RIP is a common denominator. It applies compensation to a Photoshop file, an Illustrator file or a Corel file the same way. No matter the program used, it gets the same target output. This is why it's not always best to adjust your photoshop files (within photoshop) and also have don't gain adjustments in the RIP. You would end up floating all around in your percentages and it's just harder to control.  Pick one method and stay with it.

Dot gain compensation via a RIP gets applied (across the board) to all art files (to film) or whatever end product you are outputting to. With that said, it is important to point out that inside photoshop, one standard method can be to baseline your file setup or, (keep everything the same), all the time, with no dot gain compensation built into conversions. This is so that you are always printing the compensation the same way.  As a result, you may have several QUEUES to put your art files into for a more unique compensation (based on what you are printing).

CMYK 55lpi ellipse dot shape (with each different screen angles).

Single color output (1 color prints). Maybe be more opened up to allow for more pressure for a more solid coverage in the shadow and solid areas, thus more gain in the mid tones and highlights.

Mulit color (Tee shirts)    sim process 55lpi ellipse dot shape, 22.5 degree screen angle.

Mulit color (Sweat shirts) sim process 45lpi ellipse dot shape, 22.5 degree screen angle (lower mesh).

Each may have a unique dot gain compensation plan assigned.

No RIP will do automatically. It will, after the fact, but you must enter the correct information (the compensation data) first.

To do it accurately, (as you seem to be looking to do) You must handle it yourself, by doing calibration testing on your printed garment using a reflection densitometer. There is no other way (that I know of) to achieve accuracy without one. I myself am not concerned what the output device puts on film. We should be concerned with what it puts on the printed shirt. That is the end result. That number (in the target area of reference) is where you determine where that same reference area in the curve should be adjusted to so that once printed out to your imaging device using 39% in the 50% range, you end up on the printed shirt at 49-50 or 51%. You can be happy with that.

Note, some areas in your curve may be different. You may not end up taking out 22% across the board. Some imaging devices reach a plateau where it may spike inconsistently from previous results in the curve.

It should end up printed (very close to) what you intended. (e.g.) 50%'s printed on the garment as 50%'s etc across the board.  It most often will never be 100% accurate (all the time). To do so, you would have to have all variables controlled to the letter and to have the order types consistent every time. It doesn't happen. Not even for the best of them because to do so,you become too inefficient with time and in a real world environment with high turnover of employees etc. It's just not feasible to expect perfection.  Now, with compensation, you will indeed be far more accurate than you had ever been.


Thats my story and I'm stick'n t it.
Dot-Tone-Dan

[Screen Dan]

Thank you Dottonedan.  That is an absolutely excellent post. 

I have a couple of questions.

At my shop I take care of the dot gain compensation at the RIP.  I played around with it for a while.  I did not use a densitometer.  I instead winged.  I have familiarity with all of the concepts involved...which at the core is rather simple.  My understanding of it is as such:  Regardless of the layers of complexity between the RIP and the final printed garment (and I understand the potential for dot loss and dot gain at almost every stage of the game), what was actually RIP'd versus what ends up on the garment will--in a real world environment--reflect the effective dot gain.

As such I made a number of test films ("digital films" for application via CTS) with the uncalibrated RIP.  I had those put through our (shockingly simple) process.  Except for exceptional cases (rare super fine detail or super-low mesh requiring HD inks for stacking, glitter flake, puff bases, etc) we only use two meshes.  150S and 225T at 55LPI....and on these test films I made specific designations for gray percentages on a 0-100% scale (I also did them on a 0-255 scale next to it...this actually proved more useful) and I, by eye, adjusted for our effective dot gain (~24% across the entire tonal range, give or take 5% at certain ranges...with an Ellptical-P dot there I put in ~5% dips around 30% and 70%, give or take ~5% range at each end).  Of course 0% and 100% are always going to be 0% and 100%, but I starting the adjustments at 99%....and I was shocked to find that 99% didn't look any different front 100% for, well, about 24 more percent.

Now I'm sure if we changed our mesh, ink...I'm sure different times of year this varies slightly based on local climate...and this isn't accounting for what happens in the shop.  But I'm aware that all of this can come into play.

The results, once I fine tuned this through trial and error of the course of roughly 6 months, made a staggering difference in the reliability and repeatibality of our setups...as well as decreasing setup times, requests for artwork adjustments, screens on different meshes, etc.

To me, in theory, this seems like a pretty good way to go about this...is there a way to do this better?  Am I on the right path?  What can I do to prefect this part of the process?

Also, When we first got the RIP we were encouraged to use the Elliptical-P shaped dot "because digital."  This doesn't make sense to me, but, at the time we knew nothing about how CTS operated and we went with the voodoo.  Since we're now calibrated to this dot shape--which is nominally yet measurably different from the standard 600DPI standard "Elliptical" dot I have been wondering if trying a different dot shape (like standard elliptical...or something else?) might be beneficial or more trouble than it's worth since we have a pretty effective curve going already.

Everybody is thrilled with the quality of our output and the setups have never been more reliable...but...can we do better or should we just settle down and spend our R&D days on something more worth the time?

One more:  I have started trying to come up with a custom curve just for discharge since we've noticed that dots beyond a certain percentage gain considerably more than plastisol and others clog outright and lock up pretty early in the run.  Am I barking up the wrong tree?

Thanks in advance.  I'm honored to even have the venue to ask such questions of someone as yourself.

[Dottonedan]

You had me at the edge of my seat up until you said you were honored. LOL. Thanks, but I had to chuckle a little. The gang here knows' I'm just a mere flawed mortal like everyone else.


I'm unfamiliar with the Elliptical P setting. Is this a wax or ink based CTS? I haven't heard of it, but thats not to say much.  I've learned tho, to be cautious. If they "recommend it" it may be for good reason. Still tho, what can it really hurt to try the other?  Not much. You may have to re-make a screen or two.


On both WAX and INK CTS devices, due to the WAX and INK delivery process and head configuration you do end up getting (more) dot gain at output then what you see in a preview on screen. So some "additional" compensation is required in the shadow tones on CTS before ever getting to the screen. I feel that most screen print shops far and wide just need to have some kind of data printing at that range. It doesn't need to be dead on accurate. Crikey,  , if we do ever get 95-98% dot tone in the shadows, it may or may not (easily) get obliterated just by using the wrong squeegee on press not to mention other variables. So don't sweat losing 99% at all. Shoot for 95 as your lowest and you will be doing good. Don't worry about opening up the curve enough to actually see tone (once printed on the screen). Your 95% may look light like 80% (in preview) but once it reaches the screen, gets exposed, it's another + or - a % depending...and then you have on press dot gain.


Print your test that is 0-100%  (I use .5" blocks filled with %'s from 1% to 99%).  If you can print it using black ink on a white tee and you can see dots in the 3% range and dots in the 97% range, You're doing exceptional. Don't be afraid to beef up your 1% to a 2 or 3%. That is to help you hold the smallest dot easier and enables you to get those very subtle fades to the shirt (without getting mesh interference).  Also, with that, be sure to use the proper LPI ratios to mesh.  For this area, and that type of job, (i.e) soft fades to 0%. you need to highly consider the LPI to mesh so that it does not block out the soft fades and you end up with those saw toothed jaggies.


For me, I feel if you can adjust to get some dots showing up in the 90-95% range, you are doing better than most shops.  I like to have some data in the shadows. I don't want my 80% filled in solid. I want to see deep blue shadow tones showing through my black halftones.


The discharge can have a different TYPE of gain added to the traditional or more common physical gain than plastisol mostly due to the inks becoming part of the threads (causing more of whats known as optical gain. The physical gain is there just like plastisol, but here, with discharge, those dots visually melt into the threads. Thats why so many people talk about (not seeing the dots). Check out some of the work by TonyPep and SERJ
in the this link http://www.theshirtboard.com/index.php/topic,73.msg128543/boardseen.html#new.
Do a search on their names and look up some post they have made. Most don't show any dots. They melt into the threads and I'm sure they have compensated greatly in the shadow tones for the optical gain. Think of these shadow dots as folding around the threads (molding to) the shape of the rounded threads. This will visually gain "filling in" with normal viewing distances. To show more shadow tone, you have to also add in some additional compensation "to see" some tone in those shadows. Lets say you normally cut back a 95%-98% to a 90% for plastisol. So, cut back another 5% for discharge and compare results.


Then yes, save that as a custom curve just for discharge printing.


You're certainly on the right track. I commend you for digging deep and you're at the right forum to learn all that you can to get there.

[mk162]

What I meant in "by hand" is running the calculation on the equation...not the density itself.

I have a reflection densitometer with a wide aperture that works pretty well on measuring from shirts.

The problem is the equation takes a bit, it would be easier to plug numbers into a spreadsheet or online calculator and be done with it...
http://www.brucelindbloom.com/index.html?Eqn_DotGain.html

I've always wondered if it was best to let the RIP or the graphics program handle dot gain.  I would guess RIP since you could then print from any program and it will all be handled the same way.

[Screen Dan]

Dan...thanks so much.  I had a suspicion I was on the right track...but it's nice to know.  Especially after going down this path for so long.  The ultimate measure is always in the printed results, which have been great and always improving...but it's nice to know that I've got a decent hold on the science behind it.

This is on a wax CTS.  As far as "Elliptical-P" is concerned, I was told that it is (supposedly) an Elliptical dot designed for digital...the idea being that since as dot percentage, thus size, decreases so does our ability to represent it perfectly at any given DPI.  a 70% dot looks like a dot but a 2% dot can't possibly be as perfectly shaped at the same resolution.  So it's made in a shape that scales down more accurately.  That's how I understand it.  I've experimented with it...there is no appreciable difference on a finished garment as far as precision goes.  But it does affect the curve in certain tonal ranges.  I haven't been too concerned with this so I've left it since this curve has been working pretty well for us.

As far as my focus on the 99% mark goes, I can be a bit of a perfectionist.  The current curve actually represents a very slight difference at 99% from 100%....but whether this grows or not is in the printer's hands....as well as the temperature in the shop, condition of the ink, pressure, etc etc etc.  But as tested there's a difference...which is a far cry from the uncalibrated rip.  ~80% and 100% were a solid band before.

Also, even though I cut off 4% and below to aid in repeatability (developing those dots is hard enough reliably...and then there is every other consideration before it hits the substrate that make it a hit-or-miss proposition in a high-volume shop that didn't even notice when I cut them off.) I boosted the remain low percentage dots. 

Ultimately this results in a reduced dynamic range that, aside from those 5% dips I mentioned in those two spots, but increased repeatability and decreased difficulty everywhere from developing the stencil to squeezing ink through every dot.

As far as dot angle goes I went with Murakami's recommendation for 55LPI.  Even though 225-T shouldn't work with this configuration I've found that with the steep drop off at the bottom of the curve and boosting what remains at the bottom, except in exceedingly rare circumstances, moire is never a problem.  When it is?  We've got a rack of 300-S that can handle that nicely.

Thanks for the tips man.  I can't wait for R&D Friday.

[Dottonedan]

Read this one. It's a little more clear. I think you are getting too caught up in the math equation and not seeing what you need to read from your results. The math equation must have the resulting number  of your printed target or reference area % (e.g.) 50% area from your densitometer readings for the equation to work.


Page 6 specifically.http://www.xrite.com/documents/apps/public/whitepapers/Ga00005a.pdf


You need to 0 out your shirt color (white) aka (the surface area) and then read your TINT results in the specific target areas. Here is a visual aid to assist.

[mk162]

yup, that is what ours does, except the numbers aren't clear cut like that....for instance our shirt base is -.8 and solid black is .47...

so that is where the equation comes into play, it turns those numbers into actual dot gain numbers...we calculated a couple and worked from there...a simple plug and play spreadsheet would be easier.

[ABuffington]

My 2 cents (about what it may be worth!).

Halftone control can open major doors and new markets outside of t-shirts.  The key term missing here is Lineraization, which is what we are trying to do when adjusting for dot gain.  Wasatch RIP, Filmgate from Colorgate, and many others perform all of this internally eliminating the need for PS adjustments or eye adjustments.  This is why standardizing color profiles and file info in PS works well from job to job without adjusting constantly.   (But nothing wrong with doing it by eye if the final result looks good.)  Typically in the two programs mentioned above you have a dialog box to enter in the readings from a printed tonal ramp with a transmissive densitiometer that is measuring the tonal output of the tonal steps in 5%-10% increments.  The X-rite Eye One densitometer simple reads the film tonal percentage, and is not to costly to own.  So for a 20% dot your films will read 40-45% or higher on the first go around. It is important to zero out the density with the specific film you are using. Place densitometer over film and hit zero to eliminate the 1.5-2 density of the film itself.  Entering in the measured dot allows the RIP to create a curve from the data.  It compensates the output by measuring the difference in the values of known tonal output to measured tonal output using the densistometer.  This can be an iterative process, do it a couple of times to fine tune the output.  Once film is correct then we need to look at print dot gain, which can be quite similar. 

This can be a subjective call, look at the corrected film against a print and go back and adjust the curves or input values to further control dot gain.  In Wasatch these two controls are under halftone properties>calibration.  They are called Press Curves and Calibration curves to apply the transmissive readings and the subjective print readings.
In most cases however we look at the final print to see if dot gain is killing negative dots above 60% or if we are losing dots below 10%.  Both areas can be controlled but the human eye is still needed to see if the readings created a decent print result on the shirt, (which a reflective densitometer might not read correctly, whereas it will on flat white stock for graphics.)

In most Wasatch installs I have done, ink jet gains 5-10% below 30% and above that it can gain from 20-35% depending on inkjet printer and film.  Linearization helps hold more tonal values and seems to work much better when controlled from within the RIP.  You need only enter in 5%, 10%, 20, 30, 40, 50, 60, 70, 80, 90% to get a good curve to control film dot gain imaging.  Beyond that a reflective densitometer can be used, but I have found it can also mistake fabric openings and shadow and can skew the results.  Comparing a printed tonal ramp on the fabric to the controlled film subjectively by eye eliminates the need for a $3000-6000 high end relective spectrophotometer to read 1/2 tone densities.  For tonals vales below 4%: typically the thread in the mesh is wider than the dot.  You'll image some, but other dots will get blocked leading to moire in low tonal values.  Typically visable as vignette moire on a fade into the shirt color from 100-0%.  Wasatch does have a hybrid option I have been dying to test.  You can transition from a halftone to stochaistic at the tonal value of your choosing.  Haven't dialed this in yet, just need some lab time here.

The following all affect the tonal output:  Film type, Printer brand and type, Age of printer (even identical printers they may have different linearizations due to wear and tear), Halftone line count, dot shape, printer controls on ink, distance to substrate, ink brand used, shirt to be printed on, mesh, tensions, squeegee duro, angle, speed, sharpness and pressure, ink rheology, ambient temperature, press off contact, pallet levelness and probably many more.  For t-shirt printers ball park accuracy will yield decent halftone reproduction.  For graphic printing of 4/C process every item applies.  Try printing 4/C process 60x120 movie posters in CYMK.  The control over halftones in graphics is a quantum leap in complexity of halftone controls compared to textile printing.  Graycol 7 is one method to get repeatable graphic flat stock 1/2 tone printing right and is easier than it sounds. CMY grayscale tonal ramp should match the K tonal ramp at all percentages.

For textiles the RIP to me has the best control and fastest adjustment time with less trial and error.  Some RIPs excel at this detail of control, some don't.  When you get the whole process of linearization from a quality RIP, it can allow you to output color matching proofs on your inkjet (with same halftone line count) and match to a dtg printer, and then to a large of run of shirts, to billboards and POP displays that have all been run through linearization, ICC 1440 chip color matching.  This type of color matching across many print platforms is the reason high end rips are needed to linearize all printers to run ad campaigns for major graphic users.  (Mcdonalds, Cosmetics lines, Large apparel companies.) This is where digital and screen printing can merge in color control and expand market potential for those willing to jump into high end color management and the digital print environment.  Best you can have a printer in China printing POP displays for your shirt program that match exactly with Job Definition Format files that can use printers anywhere in the world that will match precisely, which is more of a digital arena than textile, but it is being done right now by some Chinese apparel companies who are setting up shop in LA and worldwide to meet just in time deliveries, all by controlling halftone output. 

Al

[Inkworks]

That's one of the best write-ups I've read covering all the important point. linearizing your film is great, but if the post-print gains are still out of whack, then it's all for naught. Many of the points touch on important aspects some have dismissed in earlier threads.

Fact is that the finished print is where it matters most, the rest is a means to an end. Printing on fabric adds a whole level of new variables to the equation.

Very interesting info on newer rips being able to transition to stochastic dots at a chosen percentage!

[mk162]

I got it, it's super easy once I broke it down into more cells.  I was trying to put the entire formula into a single cell and pull the data from that...

Now I just have to plug my numbers into the rip...